openchat-3.5-1210-starling-slerp

<!-- header start -->

Model Description

This model uses the Slerp merge method from 2 models:

1. openchat/openchat-3.5-1210
2. berkeley-nest/Starling-LM-7B-alpha

• base model: openchat/openchat-3.5-1210

I SLERPed these two together because they're both OpenChat-ish models. Fundamentally, OpenChat-3.5-1210 appears to be trained similarly to OpenChat-3.5 but now with Feedback-Collection and a de-contaminated Capybara. Starling is OpenChat-3.5 but trained with a novel training method on the Nectar set.

My hope is that a SLERP between the two retains the benefits of both.

The yaml config file for this model is here:

slices:
  - sources:
      - model: openchat/openchat-3.5-1210 
        layer_range: [0, 32] 
      - model: berkeley-nest/Starling-LM-7B-alpha
        layer_range: [0, 32]
merge_method: slerp
base_model: openchat/openchat-3.5-1210 
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5
dtype: bfloat16